In polymer flooding, the polymer solution injectivity is a key element to estimate the success of EOR measure, which is highly related to the polymer viscosity. Because of the high shear rate near the injector wells, the polymer viscosity changes considerably due to the non-Newtonian rheological behaviour. The existing well inflow/outflow models implemented in numerical reservoir simulators are not flexible enough to consider strong shear thinning effect without building extremely fine grids. Therefore, the predicted pressure difference in the vicinity of the wells is usually overestimated.
In this paper, the way to improve the quality of the polymer injectivity calculation within the reservoir model with a coarse Cartesian grid is suggested. The idea of introducing a negative skin factor into the inflow model is extended and applied for damaged wells (in which positive skin is available in the existing model) or wells with a reduced contact surface with the formation layers due to perforation. To cover a non-uniform and high speed flow in the vicinity of discrete distributed perforation openings, special approximations for a predefined transition zone are considered. In all derivations of the analytical solutions, the power law rheology of the polymer is assumed.
The method was adapted to the commercial reservoir simulators Eclipse 100 and CMG STARS with different viscosity representative radii in the polymer inflow model. The impact of the improvements on the numerically reproduced injection of non-Newtonian fluid is tested on a synthetic box model as well as on a real field model by injection of the biopolymer Schizophyllan which exhibits a remarkable shear thinning behaviour.
Depending on the grid size resolution, the results show significant improvement in polymer injectivity prediction. Advantage of the approach becomes more distinct, as the size of the well grid block increases, which allows to significantly reduce the CPU run time.